Overfitting
Tuning a strategy until it memorizes the past instead of learning something that repeats.
What it means
Add parameters and exceptions to any strategy and its historical results improve — necessarily, because you are fitting noise. The result performs beautifully on the data it was tuned on and falls apart on data it has never seen. This is overfitting, the default failure mode of backtesting.
Defenses are procedural, not clever: fewer parameters, out-of-sample data the tuning never touched, robustness checks (does it survive ±20% on every threshold?), and deep suspicion of any result that looks too smooth.
Why traders care
- Explains most "worked in backtest, died live" stories.
- A slightly worse but stable result beats a perfect fragile one — every time.
Related terms
Educational content, not investment advice. Engine details describe how TraderWe computes this value; other platforms may define it differently.